Generating an OHDSI OMOP Data Model from HCSRN's VDW

John Weeks

Introduction

v Many of you may have heard of the OHDSI OMOP or have seen one of my presentations about their products. For those of you who are unfamiliar with OHDSI, I will begin with a brief introduction to the organization and their Common Data Model used for Healthcare Research. v Then I will discuss some of the motivations for the KP Washington Health Research Institute for creating a way to translate a selected group of patients from our VDW into an OMOP Common Data Model. v Then at a very high level we will introduce you to the process of translating between these two Healthcare Common Data models. v Finally, we will explore how you can create the OMOP common data model for yourself, by downloading the scripts that perform the translation between the VDW and the OMOP common data models. >

OHDSI (pronounced “Odyssey”) OMOP
Why Generate OHDSI OMOP from your VDW
The VDW Data Translation Process
Implementing your Translation Scripts Locally

OHDSI (pronounced “Odyssey”) OMOP Common Data Model

What is the OHDSI OMOP Common Data Model?
- A Common Data Model is a published data model that researchers can implement at their sites
- Keeps their data secure
- Allows them to share their code
  - Just like the VDW!
The Book of OHDSI is a great place to start learning about OHDSI OMOP.

Why Generate the OHDSI OMOP From Your VDW

Why is it important that the HCSRN sites be able to generate the OHDSI OMOP CDM?

Expanding Scientific Collaboration
Financial Support

Used in Projects Around the World

The OHDSI OMOP Common Data Model is used in collaborative healthcare research projects internationally.

Federal Grants Standardize Around the OHDSI OMOP

Precision Medicine Initiative: $130,000,000
21st Century Cures Act: $1,800,000,000 over 7 years

VDW Data Translation Process

The Translation of Mapped Codes
The Translation of Data

Translating Codes

The OHDSI OMOP Vocabularies provide a sane approach to UMLS.

VDW standard codes are stored in the translation software
OHDSI OMOP standard codes are stored externally in the OHDSI Athena application

Data Translation Process

In broad terms most of the VDW entities have a corresponding OHDSI OMOP entity.

SAS Scripts to Translate Your Local VDW Into OMOP

The motivation for our institute to build the OMOP came from the EMERGe project. We used our Research Data Warehouse to create a limited version of the OHDSI OMOP. That project was then modified by removing elements that were peculiar to our site and moved into a GitHub project where it became the foundation for the scripts that translate from the VDW to the OHDSI OMOP. The SAS scripts we are providing have some dependencies that will need to be met in order to generate the OMOP common data model at your site. v First, you'll obviously need an implementation of the HCSRN Virtual Data Warehouse. v Then, someone from your site will need Membership in OHDSI. v After that, your programmer will need to download a local copy of KP Washington Health Research Institute's HCSRN VDW to OHDSI OMOP SAS translation scripts. v Lastly, your programmer will need to download the OMOP Common Data Model and Vocabulary data. This will complete the pre-requisites needed to generate the OHDSI OMOP from your local VDW implementation. v

Before you can translate your VDW into OHDSI OMOP you will need:

An implementation of the HCSRN VDW
Membership in OHDSI
The KPWHRI HCSRN-VDW to OHDSI OMOP SAS translation scripts
Download OHDSI OMOP CDM and Vocabulary Data

The HCSRN VDW Common Data Model

This SAS translation script requires an HCSRN Virtual Data Warehouse (VDW) data model at your site
Information about the VDW Common Data Model is available on the HCSRN website
HCSRN membership includes access to a large collection of SAS quality assurance scripts

Membership in OHDSI

Find information on how to become a member in The Book of OHDSI

Becoming a member of the OHDSI group will provide you access to a large collection of Open Source tools and data sets
Among those OHDSI tools you will find
- OHDSI Atlas queries your OMOP common data model and saves cohort definitions
  - Many of the cohorts defined in the Phenotype Knowledge Base (PheKB) are produced by the OHDSI Atlas Application
- OHDSI Athena is an application that provides access to the OHDSI OMOP standard vocabularies

HCSRN-VDW-to-OHDSI-OMOP Translation Scripts

You can download the SAS translation scripts to generate OHDSI's OMOP from the HCSRN VDW at KP Washington Health Research Institute's organizational GitHub repository. These scripts include all of the standard codes and their descriptions that are unique to the HCSRN VDW. The scripts will also generate a set of tables that map from the standard codes in your VDW to the OHDSI OMOP code identifiers. This SAS project is open source software and freely available for download at your site right now. If your VDW doesn't implement all subject areas then you'll need to modify the translation scripts. Once implemented these SAS scripts will provide you the best of both worlds: the stable and deeply curated data of the VDW and the popular OHDSI implementation. As we work together to meet your needs the depth and quality of this SAS project will continue to improve. v

You can clone or download the scripts from VDW to OMOP GitHub webpage
These SAS scripts are published as Open Source software
They are freely available to anyone who is interested
The best of both worlds
Let's work together to expand these scripts to meet your needs

HCSRN VDW Code Sets

The SAS scripts include all of the standard codes and their descriptions that are unique to the HCSRN VDW.
During run time the scripts will generate a set of tables that map from the standard codes in your VDW to the OHDSI OMOP codes.

The HCSRN VDW Standard Codes include their descriptions, type and subject area source:

Code	Code Description	Code Type	Code Source
F	Female	SEX_ADMIN	DEMOGRAPHICS
GLU_F	Blood Sugar Fasting	TEST_TYPE	LAB_RESULTS

The OHDSI OMOP Vocabulary Data

Download the OHDSI OMOP Vocabulary by going to the OHDSI Athena website
Extract the OHDSI OMOP Vocabulary CSV files into the "omop_vocab" folder

Final Thoughts

This project will provide your organization a speedy implementation of the OHDSI OMOP common data model. Even if you don't want to build the OMOP, this project contains standardized VDW codes, code descriptions and code management scripts that have been really useful for our programmers when producing reports from our VDW. These scripts are getting regular updates that improve their functionality. You can help mature and harden this SAS project by using the software and sharing with us the results of your work or by submitting fixes or report issues directly into our GitHub repository. Thank you for watching this presentation, if you are interested in implementing the OHDSI OMOP the URL for our SAS repository is provided in the next slide. If you would like to contact me for additional information please reach out to my email address: John.M.Weeks@kp.org.

This project includes VDW standard codes and code management SAS scripts
This project is a work in progress AND you can be a part of that progress
Your feedback will help mature and harden the scripts for all of our users

Thank you for watching this presentation.

my email address is: John.M.Weeks@kp.org

List of URLs Referenced in Presentation

The URL for producing the OHDSI OMOP data model from the HCSRN VDW:
- https://github.com/kpwhri/HCSRN-VDW-to-OHDSI-OMOP
The URL for The Book of OHDSI is:
- https://ohdsi.github.io/TheBookOfOhdsi/TheBookOfOhdsi.pdf
The URL for information about HCSRN's Virtual Data Warehouse is:
- http://www.hcsrn.org/en/Tools%20&%20Materials/VDW/
The URL for the Atlas Demo is:
- http://www.ohdsi.org/web/atlas
The URL for the Athena Vocabularies application is:
- https://athena.ohdsi.org/vocabulary/list
The URL for the Phenotype Knowledge Base (PheKB) is:
- https://phekb.org/phenotypes